Data Science Seminar
September 22, 2020
Ignaty Leshchiner, PhD
Postdoctoral Fellow, Harvard Medical School/Brigham and Women’s Hospital
Zoom link: http://bit.ly/DSSept22
Real-time study of tumor emergence and progression in patients will help predict and ultimately change the course of the patient’s disease. This could be achieved by inferring genotypes of heterogeneous cell populations within the tumor, their fitness, growth rates, corresponding expression patterns and drug tolerance states. We have developed a set of computational methods to infer the order of tumor-initiating events and to follow the dynamics and competition of cancer cell populations during disease progression and treatment. The package, PhylogicNDT, uses tumor genomic data to reconstruct the process of tumor formation, natural growth kinetics, competition and spread of resistance clones. We applied this package to 2,658 primary cancers to reconstruct developmental trajectories and history of common tumor types in premalignancy and early malignancy state; reconstruct cancer cell populations and growth rates, fitness and kinetics of individual clones during natural progression of leukemia in vivo; analyze spatial progression of resistance clones and find new resistance mechanisms in a large cohort of rapid autopsy cases. By integrating blood biopsy (ctDNA), solid tissue biopsy and autopsy data we show that resistance often emerges in multiple distant metastatic sites simultaneously, with evidence of multiple resistance mutations present in the blood’s ctDNA at the same time. Finally, we combine bulk and single cell sequencing data to help identify genetically distinct clones and explain their phenotypic differences. We envision that treatment decisions will improve with better understanding of tumor development, clonal structure and microenvironment, and the path tumor takes to become malignant and progress after treatment.